The development of single-cell RNA sequencing enabled the high throughput characterization of cell populations with unprecedented detail. Yet, it failed in capturing the spatial localization of individual cells. Overcoming this, different spatial profiling methods have been developed in recent years, with in situ sequencing (ISS) being among the most powerful solutions

ISS is a targeted spatially-resolved transcriptomics method designed to detect the expression of hundreds of genes in situ in a single experiment. For this, ISS employs padlock probes, a type of oligonucleotide designed to specifically hybridize on the targeted regions, with rolling circle amplification and a combinatorial detection of the transcripts imaged. Due to its throughput and resolution, ISS is seen as a useful tool to create high content molecular maps of tissues, being of special use for building spatial atlases. However, due to its recent development, it’s still unclear how this should be done. The work presented in this thesis explores ISS as a tool for building large spatially-resolved atlases of cell types. 

In paper I, we compare the performance of cDNA-based ISS with the High Sensitivity Library Preparation Kit, developed by CARTANA AB. We identify this product to be fivefold more sensitive than cDNA-based ISS due to its improved chemistry. In addition, we show that this increased sensitivity enhances the analytical capabilities of the resulting data.    

In paper II, we build a topographic atlas of the developmental human lung. We identify 83 different cell types and states, including a novel type of GHRL-positive neuroendocrine cell. We further elucidate the developmental origin multiple populations, defining their location in situ and predicting potential interactions. 

In paper III, we create a topographic atlas of the adult human lung. We combine multiple spatial transcriptomic technologies to generate spatial maps of the populations found in the adult lung. We decipher regional differences in terms of cell type composition and cell type-specific expression. Finally, we also characterize the spatial context of rare cell types.

In paper IV, we employ large-scale data integration to construct a scRNA-seq-based cellular map of glioblastoma, an aggressive brain malignancy. In addition, we use ISS to generate single-cell resolution cell type maps of 13 glioblastoma patients, identifying consistent niches across patients and uncovering the cellular organization of these tumors. 

In paper V, we explore the quality of the data generated by the Xenium In Situ Platform, a product based on ISS and commercialized by 10X Genomics. We explore the main characteristics of the data and benchmark it against other technologies. Finally, we also define best practices for the most common analysis done using these datasets. 

Collectively, the studies presented in this thesis serve as evidence of the efficacy of ISS in constructing comprehensive cellular atlases with a single-cell resolution.

Spatially resolved transcriptomics (SRT) approaches have allowed for the investigation of transcriptomic defined cellular diversity of biological tissues previously unseen. A multitude of different SRT technologies have been developed over the years, addressing the various needs of the scientific community by enabling the characterization of molecular signatures in situ, while preserving tissue morphology.

Despite the multitude of SRT techniques developed, there is still no single ‘best’ SRT approach, due to trade offs different techniques have to compromise on. The SRT quadrilemma, termed in my thesis – Throughput, Specificity, Sensitivity and Multiplexing are the main characteristics that the dream SRT should possess, but is theoretically impossible due to the mutually exclusive nature of these characteristics. The work in this thesis focuses on the development of In Situ Sequencing (ISS) with padlock probes and rolling circle amplification, tackling the SRT quadrilemma.

In paper I, we investigate the efficiency bottlenecks of cDNA-HybISS against a commercial kit that targets RNA directly, circumventing cDNA synthesis in situ. We found that by circumventing cDNA synthesis in situ, we are able to improve the detection efficiency 5 fold. In addition, the increase in sensitivity enhanced analytical capability of our data and allowed for low, 5X magnification imaging.

In Paper II, we provide an end to end in situ transcriptomic solution with a RNA targeting ISS chemistry with improved detection efficiency compared to cDNA-HybISS and user friendly and well documented computational tools for probe design, image registration, decoding and analysis. In addition, we also demonstrate that our RNA-ISS is compatible with posterior stainings such as multiplexed antibody staining, opening up the possibility of spatial multi-omics all while maintaining cost effectiveness, customizability and ease of implementation of RNA-ISS.

In paper III, we show that we are able to achieve single nucleotide specificity with RNA targeted ISS. We show that we are able to distinguish human and mouse cells from the genotyping experiment with competing padlock probes targeting a conserved region of human and mouse beta-actin sequence that differs by a single base. In addition to the improved detection efficiency we show that the specificity with single nucleotide RNA-ISS is comparable to the established cDNA-BaSSIS method.

In paper IV, we further developed a RNA gap filling approach for genotyping. Here, we leveraged on a polymerase mediated approach for sequence capture, reverse transcribing a stretch of unknown sequences on RNA into the probe before ligation, amplification and sequencing readout. We demonstrate that we are able to fill a gap of 20nt with high fidelity as a first proof of concept experiment.

Lastly, in paper V, we employed targeted cDNA-HybISS for a proof of concept study as a high throughput molecular screening tool for a cohort study of control and schizophrenic post mortem study of the prefrontal cortex. We attempt to map cell type compositions and macroscopic tissue organization within this cohort as an exploratory study.

The work in this thesis presents the development of next generation in situ sequencing with improved sensitivity, specificity, throughput and multiplexing as a next generation molecular tool for spatial mapping of molecular signatures within biological samples in health and disease.